---
myst:
    substitutions:
        bloc1:
            python: 14-15,29
            cpp: 4,13,22
        bloc2:
            python: 18-30
            cpp: 14-26
---
# Graph "CobWeb"

For multidimensional problem, the `drawPairs` method is limited to spot correlation because of the 
way the output looks. For instance in a problem with 8 uniformly-distributed inputs and one output, 
one can get a graphic as the one shown in {numref}`dataserver_flowrate_cobweb_pairs` (obtained with 
the code below). No special trend can be seen here.

{{                                                                                                  
    "```{literalinclude} " + parent_dir                                                             
    + "/roottest/uranie/doc/dataserver/vizualisation/" + language + "/draw_flowrate_cobweb_pairs." 
    + extension + "\n"
    + ":language: " + language + "\n"
    + ":lines: " + bloc1[language] + "\n"                                                   
    + "```"                                                                                         
}}  

{{                                                                                                  
    "```{" "figure" "} " + parent_dir +                                                             
        "/roottest/build/uranie/doc/dataserver/vizualisation/" + language 
        + "/geyser_9/flowrate_cobweb_pairs.png\n"
    ":align: center\n"                                                                              
    ":name: dataserver_flowrate_cobweb_pairs\n"                                            
    + figure_scale + "\n"                                                                
    "\n"                                                                                            
    'Graphs of "drawPairs" type between the 8 uniformly-distributed inputs and the output of a 
    given problem.\n'
    "```"                                                                                           
}}

The "CobWeb" multidimensional graph, on the other hand, consists in plotting every dimension on a 
vertical axis and connecting all the points for a single event with a straight line.  It is 
particularly useful to spot correlation in high-dimension problems as it is simple to highlight a 
certain region (by changing the colour for instance) to see if the rest of the variable are randomly 
distributed or not.

{{                                                                                                  
    "```{literalinclude} " + parent_dir                                                             
    + "/roottest/uranie/doc/dataserver/vizualisation/" + language + "/draw_flowrate_cobweb." 
    + extension + "\n"
    + ":language: " + language + "\n"
    + ":lines: " + bloc2[language] + "\n"                                                   
    + "```"                                                                                         
}}  

This code for instance specifies a region-of-interest in the output, when the value are greater than 
0.97. A correlation is highlighted between these values and a region, for which the fourth input 
variable has high value while the sixth input variable values lie between 0.2 and 0.4. The result of 
this code is shown in {numref}`dataserver_flowrate_cobweb`.

{{                                                                                                  
    "```{" "figure" "} " + parent_dir +                                                             
        "/roottest/build/uranie/doc/dataserver/vizualisation/" + language 
        + "/geyser_8/flowrate_cobweb.png\n"
    ":align: center\n"                                                                              
    ":name: dataserver_flowrate_cobweb\n"                                            
    + figure_scale + "\n"                                                                
    "\n"                                                                                            
    'Graphs of "CobWeb" type between the 8 uniformly-distributed inputs and the output of a given 
    problem.\n'
    "```"                                                                                           
}}